Use mlab-ns logs to check our failure rate per-experiment

[pboothe]

It would be very handy to correlate MLab-NS requests with experiment runs, in an effort to track failure and/or abandonment rates. Then, if we saw a spike in those rates, we could fire an alert.